马红月,李温静,吴文炤,张楠,王婧.基于CEEMDAN和卷积神经网络的配电网故障选线新方法[J].电测与仪表,2024,61(10):97-103. mahongyue,Li Wenjing,Wu Wenzhao,zhangnan,wangjing.A novel fault line selection method for distribution network based on CEEMDAN and convolutional neural network[J].Electrical Measurement & Instrumentation,2024,61(10):97-103.
基于CEEMDAN和卷积神经网络的配电网故障选线新方法
A novel fault line selection method for distribution network based on CEEMDAN and convolutional neural network
A fault line selection method for distribution network based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN)and convolutional neural network (CNN) is proposed. Firstly, the CEEMDAN algorithm is used to decompose zero-sequence current signal of each line to obtain the intrinsic function of zero-sequence current of each line.Secondly, the intrinsic functions of each line are spliced together in order to obtain a time-frequency data matrix containing abandunt fault features, which corresponds to the current system operating conditions.Finally, the fault eigenvector of the time-frequency data matrix are independently mined by CNN, and the fault line selection of distribution network is realized through the fault line number output of softmax function. Simulation experiments show that the method is independent of transition resistance, detection time delay and other factors, which can accurately and effectively identify the fault lines.
